S6-EH3P: grouped reads 33161–33189 and 44100–44112 cause intermittent Modbus timeouts

[januszkadziela-lab]

Setup

• Home Assistant
• solis_modbus v4.2.6
• Solis S6-EH3P inverter
• USR-DR164 Modbus TCP ↔ RTU gateway
• 9600 / 8N1

Problem
With normal cyclic polling enabled, the integration intermittently reports:

• Exception reading input registers at 33161
• Exception reading holding registers at 44100
• No response received after 1 retries
• sometimes received pdu without a corresponding request

Network connectivity to the gateway is stable. A 300-packet ping test showed no packet loss, and TCP port 502 remains reachable.

Manual register tests

• 33161 count 2 → works
• 33161 count 29 → fails
• 33161 count 25 → works
• 33186 count 4 → works
• 44100 count 1 → works
• 44101 count 1 → fails

This suggests two separate grouped-read compatibility issues:

1. The block 33161–33189 is too large for this inverter/gateway combination. Splitting it into:
   • 33161–33185
   • 33186–33189
     resolves the timeout.
2. The Remote Dispatch telemetry block starting at 44100 also causes timeouts because register 44101 is not readable on this setup, even though 44100 itself is readable.

Local workaround tested successfully

• Split the 33161–33189 group into two polling groups.
• Skip the 44100–44112 telemetry polling group while keeping Remote Dispatch services available.

After these changes, cyclic polling and automations run without Modbus errors.

Request
Could the integration:

• split the 33161–33189 read into smaller blocks, or otherwise limit grouped read size;
• make the 44100–44112 Remote Dispatch telemetry block conditional/optional, or gracefully handle unsupported registers inside that range?

I can provide logs and exact register responses if needed.